Data-Driven Parametrization of All-Atom Force Fields for Organic Semiconductors.

Area of Science:

• Materials Science
• Computational Chemistry
• Organic Electronics

Background:

• Organic semiconductors (OSCs) are crucial for electronic devices, but their bulk properties are challenging to simulate.
• Traditional force fields (FFs) lack the necessary torsion types for comprehensive π-conjugated molecule modeling.
• Accurate molecular dynamics simulations are vital for predicting bulk properties via statistical mechanics.

Purpose of the Study:

• Introduce OSCFF, a novel force field compatible with GAFF2, designed for π-conjugated molecules.
• Enable high-accuracy prediction of molecular properties, including charge distribution and torsional profiles.
• Facilitate advanced molecular dynamics simulations for organic semiconductors.

Main Methods:

• Developed OSCFF by constructing large datasets of molecular geometries and torsion profiles.
• Utilized neural networks (NNs) for high-accuracy Restrained Electrostatic Potential (RESP) charge prediction.
• Employed automatic differentiation to fit missing dihedral parameters in GAFF2.

Main Results:

• OSCFF demonstrates high accuracy in predicting torsional energy profiles for conjugated systems.
• Achieved precise prediction of RESP charges using NN models.
• Radial distribution functions for conjugated systems were accurately reproduced.

Conclusions:

• OSCFF significantly improves the simulation accuracy of organic semiconductors.
• The developed datasets, parameters, and models are released as open-source resources.
• OSCFF is poised to advance the study of bulk properties in organic semiconductors.